Preferred Networks Releases CuPy v14 for Enhanced GPU Computing

The release of CuPy v14 signifies an advancement in Preferred Networks' open-source contributions to GPU computing.

Thursday, February 19, 2026
2 min read
Preferred Networks Newsroom
Canonical Source
Japan
Full Analysis90%
LinkedInX
What Changed

Release of CuPy v14, an open-source GPU-accelerated array library.

Source Report

Preferred Networks has released version 14 of CuPy, an open-source array library that implements a NumPy-compatible multi-dimensional array on GPUs. This update is expected to offer performance improvements and new features for developers working with GPU-accelerated scientific computing and deep learning applications, further solidifying PFN's position in the high-performance computing landscape.

Sigvera Intelligence
1CuPy v14 released with performance enhancements.
2Supports NumPy-compatible GPU array operations.
3Aims to accelerate scientific computing and AI development.
Market Impact

The release of CuPy v14 signifies an advancement in Preferred Networks' open-source contributions to GPU computing. This enhanced library can accelerate scientific research, AI development, and data analysis across various industries in APAC, potentially lowering the barrier to entry for advanced computational tasks and fostering innovation within the region's tech ecosystem.

Regional Angle

CuPy is a key tool for AI and scientific computing in APAC. Its advancements benefit researchers and developers across the region, supporting the growth of AI-driven industries and technological self-sufficiency in countries like Japan, South Korea, and Singapore.

AI & Frontier Intelligence

Where this signal fits in the broader landscape.

49 industry signalsProduct Launch
View all
View all
Verified from official source
PublisherPreferred Networks Newsroom
Publication DateFeb 19, 2026
Source TypeCompany Newsroom
Source ClassVerified Canonical
Signal Timeline
First ReportedFeb 19, 2026
IndexedMar 10, 2026
PublishedMar 10, 2026

https://www.preferred.jp/en/news/pr20260219

Read Full Source
Confidence:0.75%
Get cross-language signal intelligence

Get curated intelligence delivered to your inbox. No spam, unsubscribe anytime.

Sign in to save notes on signals.

Sign In
CompanyPreferred NetworksIndustryAI & Frontier IntelligenceRegionJapanEventProduct LaunchSourceCanonical

Stay ahead of the next signal.

Free weekly briefings with structured signal summaries. No spam, cancel anytime.